All tests fail with LOWFLOW=on
Issue #37
new
All tests fail with LOWFLOW=on, giving the error:
At line 755 of file lowflow.f90 (unit = 51)
Fortran runtime error: Cannot open file 'neo_grid.out': No such file or directory
because none of the tests provide a neo_grid.out file.
The note in Makefile above the LOWFLOW switch is
# Include higher-order terms in GK equation arising from low-flow physics
so it's not clear that we expect the tests to all have the same result with LOWFLOW=on anyway.
Comments (4)
-
-
reporter
That would be a good approach. Do we know if there's a technical reason for it being turned on by preprocessor directives? Was it @mabarnes who worked on this?
-
Yes, I did the original coding for this. I did it with preprocessor directives primarily because the ranks of some arrays (distribution-function-sized arrays) change when including these 'low flow' terms (certain terms pick up additional dependences). As this was something I did not expect users outside my orbit to use, I did not want to hardwire these rank changes for everyone else. Indeed, including these new terms introduces new physics and should be expected to change the code results.
-
reporter
Thanks Michael, that's very helpful!
- Log in to comment
Yes, I think we may well get different results with this on. It's been a while since I've looked at this, but it might be possible to switch this from a compile time option to a runtime one. This would help ensure we don't accidentally break the lowflow code and should also make it easier to explicitly test with it on/off.